This repository consists of application of Speech Denoising using DNN, CNN (1D and 2D) and RNN (LSTM) in tensorflow. I have uploaded the input audio files and also the output files from respective models. I was also able to achieve a Signal-to-Noise Ratio (SNR) value above 10 dB in all the cases which indicates that the performance of the models was pretty good.
Folder information:
DNN - Implementation of a Deep Neural Network (DNN) consisting of 4 layers with SNR value of 13.07
CNN - Implementation of a Convolutional Neural Network (CNN) using both 1D and 2D features of the audio. 1D CNN SNR - 16.28. 2D CNN SNR - 14.63.
RNN - Implementation of a Recurrent Neural Network (RNN) using a Long short-term memory (LSTM) model with SNR value of 11.19.